Title

Network hyperexcitability in a patient with partial readingepilepsy: converging evidence from magnetoencephalography,diffusion tractography, and functional magnetic resonanceimaging.

Author(s)

Fumuro, Tomoyuki; Matsumoto, Riki; Shimotake, Akihiro;Matsuhashi, Masao; Inouchi, Morito; Urayama, Shin-Ichi;Sawamoto, Nobukatsu; Fukuyama, Hidenao; Takahashi,Ryosuke; Ikeda, Akio

Citation Clinical neurophysiology (2014), 126(4): 675-681

Issue Date 2014-08-27

URL http://hdl.handle.net/2433/198809

Right

© 2014 International Federation of Clinical Neurophysiology.Licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 Internationalhttp://creativecommons.org/licenses/by-nc-nd/4.0/. NOTICE:this is the author's version of a work that was accepted forpublication in Clinical Neurophysiology. Changes resultingfrom the publishing process, such as peer review, editing,corrections, structural formatting, and other quality controlmechanisms may not be reflected in this document. Changesmay have been made to this work since it was submitted forpublication. A definitive version was subsequently published inClinical Neurophysiology, Volume 126, Issue 4, Pages675‒681, doi:10.1016/j.clinph.2014.07.033.; 許諾条件により本文ファイルは2015-08-24に公開.; この論文は出版社版でありません。引用の際には出版社版をご確認ご利用ください。This is not the published version. Please cite only thepublished version.

Type Journal Article

Textversion author

Kyoto University

1

[Title] 1

Network hyperexcitability in a patient with partial reading epilepsy: Converging 2

evidence from magnetoencephalography, diffusion tractography, and functional 3

magnetic resonance imaging 4

5

[Author names and affiliations] 6

Tomoyuki Fumuroa,b, Riki Matsumotoa,*, Akihiro Shimotakea,c, Masao 7

Matsuhashib,d, Morito Inouchic,e, Shin-ichi Urayamad, Nobukatsu Sawamotoc , 8

Hidenao Fukuyamad, Ryosuke Takahashic, Akio Ikedaa,* 9

10

aDepartment of Epilepsy, Movement Disorders and Physiology, Kyoto University 11

Graduate School of Medicine 12

54 Kawahara-cho, Shogoin, Sakyo-ku, Kyoto, 606-8507, Japan 13

bResearch and Educational Unit of Leaders for Integrated Medical System 14

54 Shogoin- kawaracho, Sakyo-ku, Kyoto, 606-8507, Japan 15

cDepartment of Neurology, Kyoto University Graduate School of Medicine 16

54 Shogoin-kawaharacho, Sakyo-ku Kyoto, 606-8507, Japan 17

dDepartment of Human Brain Research Center, Kyoto University Graduate 18

School of Medicine, Japan 19

eDepartment of Respiratory Care and Sleep Control Medicine, Kyoto University 20

Graduate School of Medicine 21

54 Kawahara-cho, Shogoin, Sakyo-ku, Kyoto, 606-8507, Japan 22

23

[*Corresponding authors] 24

2

Riki Matsumoto M.D., Ph.D. 1

Department of Epilepsy, Movement Disorders and Physiology, Kyoto University 2

Graduate School of Medicine, 54 Kawahara-cho, Shogoin, Sakyo-ku, Kyoto, 3

606-8507, Japan 4

Tel: (+81)-75-751-3662, Fax: (+81)-75-751-3663, 5

Email: matsumot@kuhp.kyoto-u.ac.jp 6

& 7

Akio Ikeda M.D., Ph.D. 8

Department of Epilepsy, Movement Disorders and Physiology, Kyoto University 9

Graduate School of Medicine, 54 Kawahara-cho, Shogoin, Sakyo-ku, Kyoto, 10

606-8507, Japan 11

Tel: (+81)-75-751-3662, Fax: (+81)-75-751-3663, 12

Email: akio@kuhp.kyoto-u.ac.jp 13

14

[Present address] 15

The present address is the same as the above one in corresponding authors. 16

17

18

Number of references 26; Number of figures 3; Number of tables 0; 19

Abstract words 269; Keywords 6; 20

21

3

Highlights 1 2

► By means of multimodal investigations, we delineated the spatial-temporal 3

characteristics of reading-induced epileptic spikes in a patient with partial 4

reading epilepsy. 5

► Katakana reading induced epileptic activation of the left posterior basal 6

temporal area. 7

► Given hyperexcitability in the whole left fronto-temporal network normally 8

recruited for reading, prolonged reading may result in epileptiform discharges 9

and clinical seizures. 10

11

12

13

Keywords 14

15

Reading Epilepsy 16

Katakana 17

Magnetoencephalography 18

Diffusion tractography 19

Functional magnetic resonance imaging 20

Japanese 21

22

4

Abstract 1 2

OBJECTIVE: 3

The pathophysiological mechanisms of partial reading epilepsy are still unclear. 4

We delineated the spatial-temporal characteristics of reading-induced epileptic 5

spikes and hemodynamic activation in a patient with partial reading epilepsy. 6

METHODS: 7

MEG was recorded during silent letter-by-letter reading, and the source of 8

reading-induced spikes was estimated using equivalent current dipole (ECD) 9

analysis. Diffusion tractography was employed to determine if the white matter 10

pathway connected spike initiation and termination sites. FMRI was employed to 11

determine the spatial pattern of hemodynamic activation elicited by reading. 12

RESULTS: 13

In 91 spike events, ECDs were clustered in the left posterior basal temporal area 14

(pBTA) during Katakana reading. In 8 of these 91 events, when the patient 15

continued to read > 30 min, another ECD cluster appeared in the left ventral 16

precentral gyrus/frontal operculum with a time-difference of ~24 ms. Probabilistic 17

diffusion tractography revealed that the long segment of the arcuate fasciculus 18

connected these two regions. FMRI conjunction analysis indicated that both 19

Katakana and Kanji reading activated the left pBTA, but Katakana activated the 20

left lateral frontal areas more extensively than Kanji. 21

CONCLUSIONS: 22

Prolonged reading of Katakana induced hyper-activation of the cortical network 23

involved in normal language function, concurrently serving as the seizure onset 24

and symptomatogenic zones. 25

5

SIGNIFICANCE: 1

Reflex epilepsy is believed to result from intrinsic hyper-excitability in the cortical 2

regions recruited during behavioral states that trigger seizures. Our case shows 3

that reading epilepsy can arise from a hyperexcitable network of cortical regions. 4

Physiological activation of this network can have cumulative effects, resulting in 5

greater reciprocal network propagation and electroclinical seizures. These 6

effects, in turn, may give insights into the brain networks recruited by reading.7

6

1. Introduction 1

2

Reading epilepsy (RE) is a type of reflex epilepsy triggered by reading and 3

usually does not involve spontaneous seizures. RE is the only reflex epilepsy 4

classified as an idiopathic localization-related epilepsy syndrome (Commission 5

on Classification and Terminology of the International League Against Epilepsy, 6

1989), but the heterogeneity of previously published observations makes this 7

classification debatable. The current proposed diagnostic scheme defines it as a 8

reflex epilepsy syndrome without specifying a generalized or focal subtype 9

(Engel, International League Against Epilepsy (ILAE), 2001). 10

Based on previous reports of patients with seizures provoked by reading, 2 11

types of RE were identified (Koutroumanidis et al., 1998): myoclonic RE and 12

partial RE. The former is characterized by myoclonic jerks of jaw without alexia 13

at seizure onset and bilateral spikes on electroencephalography (EEG). The 14

latter is rare and characterized by ictal alexia associated with a left posterior 15

temporal ictal discharge. 16

The mechanism proposed to explain triggering of myoclonic RE in 17

reading is the existence of the interaction between a hyperexcitable cortical 18

focus and a cortico-reticular loop (Radhakrishnan et al., 1995). Functional 19

imaging studies have also revealed the hyperexcitability within cortical and 20

subcortical structures (e.g., bilateral globus pallidus, left striatum, and thalamus) 21

(Archer et al., 2003; Salek-Haddadi et al., 2009; Vaudano et al., 2012). 22

Conversely, the mechanism of partial RE in reading is thought to be local cortical 23

hyperexcitability over the left posterior temporal area. This rare symptom has 24

7

been described previously in a few patients with RE, but the mechanism remains 1

elusive (Wolf, 1994; Radhakrishnan et al., 1995; Koutroumanidis et al., 1998; 2

Maillard et al., 2010). 3

The Japanese writing system has 2 distinct orthographies, Kanji 4

(morphograms) and Kana (syllabograms). Kanji characters are visual figures 5

strongly associated with semantics; thus, their pronunciations depend on the 6

context in which they appear. Kana is composed of phonological entities that are 7

somewhat comparable with the alphabets in European languages. Kana 8

orthography employs 2 visually distinct syllabaries called Hiragana and 9

Katakana; the former is usually used in combination with a Kanji and the latter is 10

used to write loanwords from European languages. Hiragana and Katakana 11

share the same lexical representations and syllabary despite different shapes. 12

Here we described a patient with partial RE provoked by reading Katakana 13

characters, but not by Kanji or Hiragana characters. A combination of different 14

noninvasive brain studies were used in the present study: 15

magnetoencephalography (MEG), which detects the dynamic propagation of 16

epileptic spikes, diffusion tensor imaging (DTI) fiber tractography, which is a 17

direct method for depicting the structural connectivity of brain network, and 18

functional magnetic resonance imaging (fMRI), which helps visualize the spatial 19

patterns of neural activity (within cortical and subcortical areas) involved in 20

reading. 21

The main purpose of this report was, in this rare case, to delineate 22

mechanisms of seizure precipitation and propagation from the viewpoints of 23

system network hyperexcitability. We also investigated the relationship between 24

8

source localization of epileptic spikes and normal functions of the reading 1

process. 2

3 2. Materials and methods 4

5

2.1. Case presentation 6

A 28-year-old right-handed male had a history of infrequent generalized 7

tonic-clonic seizures (GTCS) in the last 13 years. The first GTCS occurred at the 8

age of 15 years when the patient was at rest. Since then, 7 seizures were 9

provoked by Katakana reading. The patient described himself as follows. While 10

reading Japanese sentences comprising Katakana letters, such as while looking 11

for a particular movie title or music for several minutes in the movie rental shop, 12

he became unable to read smoothly or understand the meaning of words of 13

Katakana. When he made efforts to finish reading, his cheeks became stiff and 14

seizures occurred with loss of consciousness. 15

The brain MRI scan was normal. Prolonged video-EEG monitoring 16

showed the normal posterior dominant rhythm of 10–11 Hz and no epileptic 17

spikes at rest or while asleep. Spikes appeared in the left parieto-temporal area 18

(maximum at P3 ≧ T5, C3, frequency of ~1/20-30 s) 5 min after the patient 19

continuously read Katakana strings letter-by-letter, and not Kanji and Hiragana 20

strings. After 27 min of Katakana reading, he felt the aura and stopped reading. 21

During the aura, ictal EEG showed more frequent spikes (5/10 s) in the same 22

spatial distribution. The spikes disappeared shortly after the reading was 23

stopped. No motor manifestations or other behavioral changes were associated 24

with spikes on video-EEG. On the other hand, paragraph reading (containing 25

9

Kanji, Katakana, and Hiragana) did not provoke any spikes. Finding of specific 1

Hangul words from random-aligned Hangul letters did not elicit any spikes either, 2

where the patient had no experience of Hangul scripts reading. Hangul is 3

comparable with the alphabets in European languages. 4

5

2.2. MEG, EEG, and MRI acquisition 6

MEG examination was performed for the spike foci. Informed consent was 7

obtained from him. MEG was recorded with a 306-channel whole-head MEG 8

system (Neuromag, Helsinki, Finland) in a magnetically shielded chamber. EEG 9

was simultaneously obtained using 21 scalp electrodes according to the 10

International 10–20 system. The sampling rate was 1500 Hz. MEG and EEG 11

data were digitally filtered at a bandpass width of 0.1–400 Hz. MRI scans were 12

performed after MEG acquisition. Diffusion-weighted images (DWI), fMRI, and a 13

T1-weighted anatomical image were acquired on a 3-Tesla Trio scanner 14

(Siemens, Erlangen, Germany). The parameters of DWI, fMRI, a T1-weighted 15

image, and the dual-gradient field map have been reported previously (Oguri et 16

al., 2013). 17

18

2.3. Reading tasks during MEG recording 19

During MEG recording, the patient performed a set of tasks: the patient sat in a 20

chair placed approximately 100 inches from a computer screen. The tasks were 21

designed to simulate the situation where he had provoked seizures during 22

reading many Katakana titles/names on the CD or DVD covers in the rental shop. 23

The screen showed strings of 600‒800 letters of Katakana. These were 24

10

essentially randomized strings without any association, and among them, 1

several meaningful words (e.g., foreign singer’s names) were intermixed on a 2

page screen (Fig. 1). The words employed as test materials were highly 3

common in daily life. We instructed the patient to read every letter from left to 4

right in each line covertly. The patient read lines from the top to the bottom of a 5

page. The patient was asked to identify real words among the strings and speak 6

loudly to the examiner once he read over the strings of letters on the screen. It 7

made the patient carefully read each Katakana letter on the screen. As controls, 8

the same tasks were performed with Kanji or Hiragana. As another control, the 9

task of finding specific strings among random-aligned marks was performed. 1 10

session consisted of 1 page of a screen with Hiragana, Kanji, or marks. Only the 11

Katakana task was repeated for 10 pages of screen in a session. Each session 12

was repeated twice. The patient performed each task randomly in the following 13

order: (1) Hiragana, (2) Kanji, (3) Katakana, (4) break, (5) mark, (6) Kanji, (7) 14

mark, (8) Katakana, and (9) Hiragana. 15

16

2.4. Generator sources of MEG spikes 17

MEG spikes provoked by the aforementioned Katakana reading and control 18

tasks were visually inspected. Equivalent current dipoles (ECDs) were 19

calculated for spikes using a single sphere model. We followed the analysis 20

procedure described previously (Enatsu et al., 2008). In order to better delineate 21

the anatomical localization of ECDs, ECDs identified on the T1 volume 22

acquisition (3T, MPRAGE) were non-linearly co-registered to the Montreal 23

Neurological Institute (MNI) standard space (ICBM-152) using FNIRT of the FSL 24

11

version 4.1.2 software (www.fmrib.ox.ac.uk/fsl/fnirt/). This method has been 1

reported elsewhere for standardization of electrode locations (Matsumoto et al., 2

2012). 3

4

2.5. Diffusion tractography 5

For tractography, we employed DWI data to trace the patient’s white matter 6

pathway between the 2 regions shown by clustered ECDs in the MEG study. 7

Probabilistic diffusion tractography was performed on the basis of the 2 regions 8

of interest (ROIs)-based approach. The details of this method have been 9

reported previously (Oguri et al., 2013). For the seed and target points, each 10

ROI was drawn as a sphere located in averaged coordinates of clustered ECDs, 11

as in a previous study (Kamada et al., 2007). In order to exclude the error course, 12

such as the tracts into the contralateral hemisphere or the cortico-spinal tract, 13

exclusion ROIs were obtained at the cerebrospinal fluid and midline. One ROI 14

was set as a seed image and the other was waypoint and termination mask. This 15

procedure was conducted in both directions between the 2 ROIs. The results for 16

each track were combined and thresholded at 10% of the maximum connectivity 17

value. Next, the tract was binarized and smoothened with a 1-mm full-width half 18

maximum Gaussian kernel for 3D display in the patient’s MPRAGE. 19

20

2.6. FMRI 21

Three types of visual stimuli such as (i) Katakana (4 characters), (ii) Kanji (2 22

characters), and (iii) the control script (Tibetan, 2 characters totally different from 23

Kana or Kanji) were displayed. To equate the retinal image size of the stimuli 24

12

among every script form, we placed an asterisk (*) at the beginning and the end 1

of each Kanji word and Tibetan scripts. Katakana and Kanji words were matched 2

for sound and meaning. Familiarity values of all employed words were high, 3

being above 5.00 according to the 7-point rating scale in Japanese (Amano and 4

Kondo, 1999). After the removal of any possible characters resembling the 5

Katakana and Kanji characters used in the present study, Tibetan script was 6

used as the control script. Since the patient had not learnt Tibetan, the letter 7

strings of this language were completely unfamiliar and provided no linguistic 8

information (i.e., word sound and word meaning). 9

The block design was done with alternating 24-s task blocks and 3-s rest 10

blocks. In each task block, a small fixation cross mark appeared at the center of 11

the visual display, and 16 words were presented at a rate of 1/1.5 s, with 300 12

msec of display duration followed by 1200-msec blank period. Sixteen 13

consecutive words in 1 task block belonged to the same category of each script. 14

Each script condition was executed in the order of Katakana, Kanji, and the 15

control script. During the rest blocks, only the fixation cross appeared on the 16

screen. This procedure was repeated 4 times with a rest intersession of about 1‒17

2 min. The patient was instructed to fix his gaze at the center of the screen in 18

each trial. In the Katakana and Kanji blocks, he covertly repeated each word in 19

his mind, and pressed the button using his right index finger when the shown 20

words were judged as food names. In the Tibetan block, he pressed the button 21

when the same 2 letters appeared serially. Within each block of 16 words, there 22

were 3 to 5 occasions on which the patient had to respond. 23

13

Before fMRI recording, he had a test run outside the scanner to ensure 1

that task words were familiar to him and that no EEG spikes were induced by the 2

task. To prevent word-specific practice effects, the task words in the test run 3

were entirely different from those used in the real fMRI recording. Hangul scripts 4

were used as the control in the test run. 5

Statistical analyses of his performance in terms of the accuracy and 6

reaction time across each script condition were performed with SPSS (version 7

15.0j, SPSS Inc., Chicago, Illinois, USA). The regional blood oxygen level 8

dependent (BOLD) effect images were co-registered into his MPRAGE. All 9

imaging procedures and statistical analyses were completed using the FSL and 10

SPM version 8 software (Welcome Department of Cognitive Neurology, London, 11

UK). We conducted subtraction analysis between Katakana and control, and 12

between Kanji and control. Conjunction analysis was performed by combining 13

the 2 subtraction data. The subtraction between Katakana and Kanji was aimed 14

to identify the activated regions associated with the Katakana or Kanji effect. The 15

subsets of voxels exceeding a threshold of P < 0.001 (Z > 3.11) without 16

correction for multiple comparison were considered to be significant. 17

18

3. Results 19

3.1. Spikes triggered by Katakana reading 20

The covert letter-by-letter reading of Katakana evoked many MEG spikes that 21

fulfilled the ECD criteria (103 times /3890 s) (the number of spikes/task duration 22

in total of 2 sessions), whereas that of Hiragana and Kanji and the task of finding 23

specific strings among the random-aligned marks provoked fewer ones 24

14

(Hiragana: 3 times /641 s, Kanji: 5 times /585 s, finding specific strings: 1 time 1

/358 s). In Katakana reading, 91 ECDs out of 103 spike-complex were calculated 2

as a single ECD and clustered in the posterior part of the left basal temporal 3

area (pBTA). The averaged MNI coordinate of 91 ECDs was located at (x, y, z = 4

−44, −46, −26). More spikes were provoked in the second session (73) than in 5

the first session (18). 6

In 8 out of 91 ECDs, an additional ECD was estimated in the left ventral 7

PrCG/frontal operculum with a little time difference to the one at the left pBTA. 8

All 8 pairs of ECDs were detected during the second Katakana reading session 9

when the patient performed Katakana reading for a total of more than 36 min. 10

ECD at the left ventral PrCG/frontal operculum preceded the one at the left 11

pBTA by 24 ± 7 (mean ± standard deviation) msec (Fig. 2). 12

13

3.2. Spike propagation tract 14

The above results suggested the existence of underlying anatomical white 15

matter pathway between the 2 foci. The left ventral PrCG/frontal operculum and 16

pBTA were traced by probabilistic tractography. The tract ran through the long 17

segment of the arcuate fasciculus (Fig. 3) (Catani et al., 2012; Martino et al., 18

2013). 19

20

3.3. Brain network for Katakana reading 21

Under the scanner, the performances of the accuracy for each script conditions 22

were 90, 90, and 92% for the Katakana, Kanji, and control conditions, 23

respectively. There was no significant statistical difference among them 24

15

[One-way ANOVA, F(2, 47.12) = 0.06, P = 0.94]. Similarly, the mean reaction 1

times did not significantly differ among the conditions of Katakana (0.80 s ± 0.14; 2

mean ± standard deviation), Kanji (0.82 s ± 0.10), and control (0.81 s ± 0.11) 3

[One-way Anova, F(2, 0.21) = 0.06, P = 0.94] in the practice run. 4

Figure 4A illustrates the representative BOLD signal changes in the left ventral 5

PrCG/frontal operculum (Fig. 4A: Right) and left pBTA (Fig. 4A: Left), 6

respectively. In contrast to control scripts, the Katakana and Kanji words that 7

were meaningful for the patient significantly activated the left pBTA (peak Z 8

values: Katakana = 5.50, Kanji = 7.32) and ventral PrCG/frontal operculum 9

(Katakana = 8.26, Kanji = 5.04) (figure not shown). More importantly, 10

conjunction analysis of Katakana and Kanji words revealed significant activation 11

in the left pBTA (peak Z value = 5.47: Fig. 4A: Left) at around the spike focus 12

induced by Katakana reading. When activation during Katakana and Kanji word 13

reading was compared by subtraction, the left lateral ventral frontal area 14

including the spike focus showed greater activation for Katakana than Kanji (the 15

“Katakana effect”, peak Z value = 7.62: Fig. 4A: Right), but this effect was not 16

present in pBTA. On the other hand, no increased activation for Kanji over 17

Katakana was observed in both the regions (“Kanji effect”). 18

19 4. Discussion 20

21

The patient was diagnosed as partial RE because of 1) alexia and stiffness in his 22

cheeks and 2) regional EEG and MEG spikes seen both interictally and during 23

the aura. Absence of the oral myoclonus and generalized or bilateral spikes 24

supports the diagnosis of partial RE. Although ictal EEG patterns were not 25

16

recorded except for that of aura, clustered ECDs likely indicate that the seizure 1

onset zone or the primary epileptic focus was located in the left posterior basal 2

temporal area. Based on his clinical findings, the symptomatogenic zone is 3

presumed to include both the left ventral frontal area and left posterior basal 4

temporal area. 5

The aim of this study was to investigate the mechanism of partial RE. 6

Previously, the hyperexcitable zone has been thought to be restricted to the left 7

hemispheric cortical region related to the posterior language area in patients 8

with partial RE. However, this study clearly showed the scientific evidence of 9

network hyperexcitability between the 2 distinct cortices and that linked 10

substructures within the left hemisphere contribute toward the development of 11

epileptic syndromes in a partial RE. 12

It is currently recognized that the orthographic information is processed 13

in the left pBTA (Brodomann area 37) (Sakurai et al., 2008). In our study, the 14

averaged coordinates of spike ECDs clustered in the left pBTA were situated 15

close to the regions required for the overt Kanji word reading (Sakurai et al., 16

2000), covert pseudo word reading (Cappa et al., 1998), and the visual word 17

form area (VWFA) (Jobard et al., 2003) (Fig. 4B). Conjunction analysis showed 18

that the reading of Katakana and Kanji words activated the left pBTA overlapping 19

or adjacent to the location of these ECDs (Fig. 4A: Left). These results suggest 20

that the primary epileptogenic area played an important role in word recognition 21

through morphological processing in our patient. 22

MEG study showed that the patient had an additional epileptogenicity in 23

the left ventral PrCG/frontal operculum. This area was overlapping or adjacent to 24

17

the activated regions of the Katakana effects in fMRI study (Fig. 4A: Right). 1

Recent functional imaging and lesion studies revealed that left lateral frontal 2

areas, namely, the ventral PrCG and frontal operculum, are involved in 3

articulation processing (Baldo et al., 2011; Price et al., 2003). In addition, the 4

syllabic character of Kana processing is more strongly involved in phonological 5

conversion and articulation as compared with Kanji processing (Thuy et al., 6

2004). These findings suggested that the significant Katakana effect observed in 7

our patient is predominantly associated with increased demands of phonological 8

conversion and articulation processing during Katakana reading. 9

Our findings suggested that as proposed in myoclonic RE by previous 10

studies (Ferlazzo et al., 2005), the cerebral networks subserving epileptic activity 11

in partial RE comprise areas of the brain involved in normal articulation and 12

morphological recognition processing. 13

It was the second session of Katakana letter-by-letter reading (reading 14

more than 30 min) that provoked considerably more spikes in the left pBTA. 15

Moreover, only the second session generated epileptic spikes in the left 16

ventralPrCG/frontal operculum. Although there were only 8 pairs of ECDs 17

localized in the left ventral PrCG/frontal operculum and left pBTA, this interesting 18

observation led us to hypothesize the following mechanism for generation of the 19

clinical epileptic seizure: 20

1) There is considerable evidence for normal activation in the cortical areas of 21

left pBTA and ventral PrCG/frontal operculum during reading (Taylor et al., 22

2013). Katakana reading network was very close to or overlapping with seizure 23

onset and symptomatogenic zone. 24

18

2) Prolonged Katakana reading provoked epileptic activation in the left pBTA, 1

the presumed primary focus. Given hyperexcitablity in the reading network led to 2

the spike propagation to the left ventral PrCG/frontal operculum via the long 3

segment of the arcuate fasciculi. 4

3) Reciprocal neuronal excitation or overload of a critical mass of neurons within 5

the left fronto-temporal reading network might have further enhanced normal 6

physiological activation, then epileptic activity and finally provoked seizures. 7

Depending on the level of the network hyperexcitability, the cumulative 8

effect of more factors may be needed for generation of the paroxysmal response, 9

namely, 1) comprehensive complexity inherent in orthography and 2) letter 10

familiarity. First, the difficulties in reading Katakana can be explained by the 11

feature model of cognitive psychology: strokes can be considered as features 12

that distinguish 1 letter from another. The letter comprising more strokes has 13

more distinctive features. The mean number of strokes in Katakana is less than 14

that in Hiragana and Tibetan, and is almost one-quarter of that in Kanji. 15

Moreover, Katakana has the most angular orthography, whereas Hiragana has 16

the most cursive orthography, and Katakana is prescribed by the position and 17

direction of a stroke (as inン (n) and ソ (so)). Hence, the Katakana, which 18

comprises fewer distinctive features, needs more effort to be read than both 19

Kanji and Hiragana. This assumption is in agreement with the finding that only 20

Katakana reading with higher degree of cognitive difficulty produced epileptic 21

spikes in the left pBTA during prolonged reading. Second, of the 3 types of 22

written Japanese characters, Katakana letters are less familiar than Kanji and 23

Hiragana, because they are mainly used for words that have been imported from 24

19

foreign languages. 1

Our MEG study showed that Katakana reading provoked considerably 2

more spikes in the second session than in the first one, and all 8 pairs of ECDs 3

localized in the left ventral PrCG/frontal operculum and left pBTA were recorded 4

in the second session. These findings suggest that prolonged effort to sustained 5

concentration caused an accumulation effect, which accelerated the excitability 6

in the above spike foci and arcuate fasciculus. This idea supports the hypothesis 7

that, under intrinsic predisposition of network hyper-excitability, the reciprocal 8

neural excitation between the two spike foci contributed to clinical seizures in 9

partial RE. 10

Spikes in the left ventral PrCG/frontal operculum preceded those in the 11

left pBTA by approximately 24 msec, a reasonable time difference for neural 12

transmission through the arcuate fasciculus according to the study of 13

cortico-cortical evoked potentials (Matsumoto et al., 2004). It is not exactly 14

known why spikes in the ventral PrCG/frontal operculum occurred first. It may be 15

something intrinsic to this patient's frontal-temporal network connectivity, i.e. the 16

higher likelihood of reciprocal neural excitation, that predisposed the patient to 17

transformation of normal physiological activation into epileptic one. Reciprocal 18

neural excitation or, if present, feed-forward and backward loops within this 19

network might account for generation of the preceding spikes in the ventral 20

frontal area. 21

22

5. Conclusions 23

24

20

Our results indicate that 1) selective subsystems including articulation and 1

morphological processing served as the seizure onset and symptomatogenic 2

zones and 2) network hyperexcitability within the left hemisphere contributed to 3

the development of clinical seizures in a patient with partial RE. In conclusion, 4

this multimodal case study implicated that, under certain predisposition of 5

network hyper-excitability, prolonged reading could enhance physiological 6

network activity and then epileptic activity, and finally epileptic seizures 7

occurred. 8

9

21

Acknowledgements 1

2

This work was supported by Grants-in-Aid for Scientific Research (B) 26282218, 3

26293209, (C) 26330175 from the Ministry of Education, Culture, Sports, 4

Science and Technology of Japan (MEXT), and the Research Grants from the 5

Japan Epilepsy Research Foundation. 6

7

Conflict of interest 8

9

The authors declare no competing financial interests. 10

11

22

References 1

2

Amano S, Kondo T. Japanese NTT database series: Lexical properties of 3

Japanese (I). Tokyo: Sanseido, 1999. 4

Archer JS, Briellmann RS, Syngeniotis A, Abbott DF, Jackson GD. 5

Spike-triggered fMRI in reading epilepsy: Involvement of left frontal cortex 6

working memory area. Neurology 2003;60:415-421. 7

Baldo JV, Wilkins DP, Ogar J, Willock S, Dronkers NF. Role of the precentral 8

gyrus of the insula in complex articulation. Cortex 2011;47:800-807. 9

Cappa SF, Perani D, Schnur T, Tettamanti M, Fazio F. The effects of semantic 10

category and knowledge type on lexical-semantic access: A PET study. 11

NeuroImage 1998;8:350-359. 12

Catani M, Dell'Acqua F, Bizzi A, Forkel SJ, Williams SC, Simmons A, et al. 13

Beyond cortical localization in clinico-anatomical correlation. Cortex 14

2012;48:1262-1287. 15

Commission on Classification and Terminology of the International League 16

Against Epilepsy. Proposal for revised classification of epilepsies and 17

epileptic syndromes. commission on classification and terminology of the 18

international league against epilepsy. Epilepsia 1989;30:389-399. 19

Enatsu R, Mikuni N, Usui K, Matsubayashi J, Taki J, Begum T, et al. Hashimoto 20

N. Usefulness of MEG magnetometer for spike detection in patients with 21

mesial temporal epileptic focus. NeuroImage 2008;41:1206-1219. 22

Engel J,Jr, International League Against Epilepsy (ILAE). A proposed diagnostic 23

scheme for people with epileptic seizures and with epilepsy: Report of the 24

23

ILAE task force on classification and terminology. Epilepsia 1

2001;42:796-803. 2

Ferlazzo E, Zifkin BG, Andermann E, Andermann F. Cortical triggers in 3

generalized reflex seizures and epilepsies. Brain 2005;128:700-710. 4

Jobard G, Crivello F, Tzourio-Mazoyer N. Evaluation of the dual route theory of 5

reading: A metanalysis of 35 neuroimaging studies. NeuroImage 6

2003;20:693-712. 7

Kamada K, Todo T, Masutani Y, Aoki S, Ino K, Morita A, et al. Visualization of 8

the frontotemporal language fibers by tractography combined with functional 9

magnetic resonance imaging and magnetoencephalography. J Neurosurg 10

2007;106:90-98. 11

Koutroumanidis M, Koepp MJ, Richardson MP, Camfield C, Agathonikou A, Ried 12

S, et al. The variants of reading epilepsy. A clinical and video-EEG study of 13

17 patients with reading-induced seizures. Brain 1998;121:1409-1427. 14

Maillard L, Vignal JP, Raffo E, Vespignani H. Bitemporal form of partial reading 15

epilepsy: Further evidence for an idiopathic localization-related syndrome. 16

Epilepsia 2010;51:165-169. 17

Martino J, De Witt Hamer PC, Berger MS, Lawton MT, Arnold CM, de Lucas EM, 18

et al. Analysis of the subcomponents and cortical terminations of the 19

perisylvian superior longitudinal fasciculus: A fiber dissection and DTI 20

tractography study. Brain Struct Funct 2013;218:105-121. 21

Matsumoto R, Nair DR, LaPresto E, Najm I, Bingaman W, Shibasaki H, et al. 22

Functional connectivity in the human language system: A cortico-cortical 23

evoked potential study. Brain 2004;127:2316-2330. 24

24

Matsumoto R, Nair DR, Ikeda A, Fumuro T, Lapresto E, Mikuni N, et al. 1

Parieto-frontal network in humans studied by cortico-cortical evoked 2

potential. Hum Brain Mapp 2012;33:2856-2872. 3

Oguri T, Sawamoto N, Tabu H, Urayama S, Matsuhashi M, Matsukawa N, et al. 4

Overlapping connections within the motor cortico-basal ganglia circuit: 5

FMRI-tractography analysis. NeuroImage 2013;78:353-362. 6

Price CJ, Gorno-Tempini ML, Graham KS, Biggio N, Mechelli A, Patterson K, et 7

al. Normal and pathological reading: converging data from lesion and 8

imaging studies. NeuroImage 2003;20 Suppl 1:S30-41. 9

Radhakrishnan K, Silbert PL, Klass DW. Reading epilepsy. An appraisal of 20 10

patients diagnosed at the mayo clinic, rochester, minnesota, between 1949 11

and 1989, and delineation of the epileptic syndrome. Brain 1995;118:75-89. 12

Sakurai Y, Momose T, Iwata M, Sudo Y, Ohtomo K, Kanazawa I. Different 13

cortical activity in reading of kanji words, kana words and kana nonwords. 14

Brain Res Cogn Brain Res 2000;9:111-115. 15

Sakurai Y, Terao Y, Ichikawa Y, Ohtsu H, Momose T, Tsuji S, et al. Pure alexia 16

for kana. characterization of alexia with lesions of the inferior occipital cortex. 17

J Neurol Sci 2008;268:48-59. 18

Salek-Haddadi A, Mayer T, Hamandi K, Symms M, Josephs O, Fluegel D, et al. 19

Imaging seizure activity: A combined EEG/EMG-fMRI study in reading 20

epilepsy. Epilepsia 2009;50:256-264. 21

Taylor JS, Rastle K, Davis MH. Can cognitive models explain brain activation 22

during word and pseudoword reading? A meta-analysis of 36 neuroimaging 23

studies. Psychol Bull 2013;139:766-791. 24

25

Thuy DH, Matsuo K, Nakamura K, Toma K, Oga T, Nakai T, et al. Implicit and 1

explicit processing of kanji and kana words and non-words studied with fMRI. 2

NeuroImage 2004;23:878-889. 3

Vaudano AE, Carmichael DW, Salek-Haddadi A, Rampp S, Stefan H, Lemieux L, 4

et al. Networks involved in seizure initiation. A reading epilepsy case studied 5

with EEG-fMRI and MEG. Neurology 2012;79:249-253. 6

Wolf P. Epileptic seizures and syndromes: With some of their theoretical 7

implications. London: John Libbey, 1994. 8

9

10

26

Figure legends 1

Fig. 1 2

An example of the screen with randomized strings of Katakana letters. The 3

patient was instructed to read every letter covertly and identify meaningful words 4

among the strings. “ビートルズ” is a Katakana word of the famous singer group 5

“Beatles”. 6

7

Fig. 2 8

(A) MEG spikes and their gradiometer contour maps. Two enlarged MEG spikes 9

with a time difference of 23 msec are clearly identifiable. Each gradiometer 10

contour map indicates 1 dipole in the left hemisphere. 11

(B) Two different ECD foci were estimated from a spike-complex: one in the left 12

ventral PrCG/frontal operculum (upper) and the other in the left inferior temporal 13

area (lower) (shown on T1-weighted sagittal MRI slices). ECD in the left ventral 14

PrCG/frontal operculum preceded that in the left pBTA by 23 msec. 15

16

Fig. 3 17

(A) T1-weighted sagittal slices showing clusters of ECDs located in the left 18

ventral PrCG/frontal operculum (upper) and left pBTA (lower). 19

(B) Three-dimensional reconstructions of functional information including the 2 20

ROIs (blue) and the result of probabilistic tractography (red). 21

The white matter fiber tract that links the 2 ROIs via a dorsal projection arching 22

around the Sylvian fissure is thought to be the long segment of the arcuate 23

fasciculus. 24

27

1

Fig. 4 2

(A) The mean location of paired spike ECDs (red circles) and brain regions 3

associated with conjunction analysis of Katakana and Kanji > control (blue 4

areas: left column) and Katakana > Kanji (yellow areas: right column) are shown. 5

Left column: Overlap of the two loci suggests that left pBTA plays an important 6

role in word recognition through morphological recognition processing. 7

Right column: The significant activation observed in left ventral PrCG/frontal 8

operculum may be predominantly associated with increased attention due to the 9

demands of phonological conversion and articulation processing. 10

(B) MNI coordinates in the present and previous studies. 11

The mean coordinates of spike ECDs detected in left pBTA in this study (red 12

circle) were situated close to the regions required for pseudoword encoding 13

(blue circle: Cappa et al., 1998), reading Kanji words aloud (yellow circle: 14

Sakurai et al., 2000), and visual word form area (VWFA) (green circle: Jobard et 15

al., 2003), but not for the covert reading of Kana words (white circle: Sakurai et 16

al., 2001). 17

*Cappa et al., 1998 18

**Sakurai et al., 2000 19

***Jobard et al., 2003 20

****Sakurai et al., 2001 21